Raman Spectroscopic Study of TiO₂ Nanoparticles’ Effects on the Hemoglobin State in Individual Red Blood Cells

Abstract

Titanium dioxide (TiO₂) is considered to be a nontoxic material and is widely used in a number of everyday products, such as sunscreen. TiO₂ nanoparticles (NP) are also considered as prospective agents for photodynamic therapy and drug delivery. These applications require an understanding of the potential effects of TiO₂ on the blood system and its components upon administration. In the presented work, we analyze the interaction of TiO₂ nanoparticles of different crystal phases (anatase and rutile) with individual rat Red Blood Cells (RBC) and the TiO₂ influence on the oxygenation state and functionality of RBC, estimated via analysis of Raman spectra of Hemoglobin (Hb) and their distribution along individual RBC. Raman spectral signals also allow localization of the TiO₂ NP on the RBC. No penetration of the NP inside RBC was observed; however, both kinds of TiO₂ NP adsorbed on the RBC membrane can affect the Hb state. Mechanisms involving the NP–membrane–Hb interaction, resulting in partial deoxygenation of Hb and TiO₂ photothermal effect on Hb under Raman laser excitation, are suggested. The possible influence on the safety of TiO₂ use in advanced medical application, especially on the safety and efficiency of photothermal therapy, is discussed.

1. Introduction

Current perspectives on developing nanoparticles’ (NP) applications for theranostics [1,2] require an understanding of NP interaction with, and potential effects on, the blood system and its components upon administration via injection. NP can also enter the blood stream after administration through breath, dermal deposition, and absorption in the gastrointestinal system. The interaction of various NP with blood and blood components has been studied. The state and functionality of the blood is determined in a significant degree by the state and functionality of red blood cells (RBC). Thus, one of the important criteria for NP use is how the NP can interact with RBC and affect the RBC functionalities. The influence of different kinds of NP on RBC under various conditions was studied, including carbon [3,4,5], silica [6,7], gold and silver [8,9,10,11], Fe₃O₄ [12] NP, and others [8]; mechanisms of interaction and the conditions of safe application were discussed. Although among other NP, titanium dioxide (TiO₂) is often considered as a nontoxic material and is widely used in everyday products of pharmaceutics, cosmetics, and food industries, the problem of its safety when interacting with blood becomes relevant due to current research concerning applications of TiO₂ NP for photodynamic therapy and drug delivery [13].

The interaction of TiO₂ with RBC has been studied in a number of works [5,8,14,15,16,17,18], and in most of them different kinds of damaging effects were observed, from morphology abnormalities [17], hemolysis [14], membrane properties modification [18], and increasing procoagulant activity of RBC [15], up to induced cell death, genotoxicity, and inflammation [5,19], which depended on internalization pathways, ROS production, and the abnormal interaction of proteins [5]. The effect of TiO₂ NP on blood was also observed in in vivo studies [20]. Therefore, TiO₂ NP exhibit potential toxicity, and an understanding is needed of how and under which conditions TiO₂ can be used to avoid fatal effects or, on the other hand, to induce toxicity for nanomedical purposes.

The protein hemoglobin (Hb) contained in RBC is responsible for the function of blood gas transportation. Violations or modifications of this function by NP can be observed [3,4,10] and need to be controlled. To investigate these effects, noncontact and noninvasive means are essential. For this purpose, Raman spectroscopy is a suitable tool for analysis of the Hb function and its dynamics due to a selective enhancement of the Raman active vibrations in different forms of Hb [3,4,10,21,22,23]. Additionally, Raman mapping allows for a direct observation of the distribution of Hb in different states. It has been applied for analyzing the formation of Hb derivatives in RBC under varying conditions [21,24], studying the process of erythrophagocytosis of senescent RBC [23], and for diagnosing malaria [24].

Raman spectroscopy and Raman mapping were shown to be effective tools for analysis of NP–RBC interaction. Previously, the effects of NP on the oxygenation state of individual RBC were studied for nanodiamond (ND), and it was shown that the adhesion of ND on the RBC membrane affects the cell functioning [3,4,18], but safe conditions for applications can be selected. NDs of different sizes stick to the RBC membrane, slightly affect the deoxygenation degree of RBC, and influence the oxygenation–deoxygenation dynamics of RBC in size-, concentration-, and surface-chemistry-dependent ways [3,4]. More serious disturbance effects were observed in the study of Au and Ag NP interaction with RBC [10]. Increasing hemoglobin deoxygenation was observed in both Au and Ag NP-treated RBC. These studies suggest that the adhesion of NP on the cell membrane causes an imbalance in RBC functioning and cell damage [10,11]. Independent of size and chemical composition, NP penetration into RBC was not observed in these works. In contrast to this, the penetration through the RBC membrane of NP and their aggregates with sizes less than 200 nm was observed using fluorescence microscopy for polymer (with fluorescent dye) NP and via electron microscopy for gold and TiO₂ anatase NP after incubation for 4–48 h [8]. Larger particles or aggregates were found stuck to the RBC surface.

In the present work, we analyze the interaction of TiO₂ NP of different crystal phases (anatase and rutile) with individual rat RBC. The oxygenation state and functionality of RBC is estimated via analysis of Raman spectra from the Hb Raman mapping of an individual RBC. Raman mapping allows observation of the distribution of Hb in different oxygenation states in the studied RBC. Simultaneously, Raman spectra allow localization of the TiO₂ NP on the RBC. In this work, no penetration of the NP inside the RBC was observed; however, both kinds of TiO₂ NP were adsorbed on the RBC membrane, which may alter the oxygenation degree in the conditions when RBC should be oxygenated or, in general, could affect the Hb form. The effects of rutile and anatase TiO₂ NP are compared. Possible mechanisms of influence of the membrane-localized TiO₂ NP on Hb are discussed.

2. Materials and Methods

To estimate the effect of TiO₂ nanoparticles on the RBC oxy/deoxy state, two kinds of TiO₂ NP with predominant rutile, TiO₂(r) (Sigma, Saint Louis, MO, USA), and anatase, TiO₂(a), (UV100 Hombikat, Sachtleben Chemie GmbH, Duisburg, Germany) structures were used. The nominal size of TiO₂(r) particles was ≤100 nm, and that of TiO₂(a) was about 10 nm. In the experiments, the particles’ size and ξ-potential were estimated using the dynamic light scattering method (DLS) with a Zetasizer Nano ZS (Malvern Instruments, Malvern, UK). NP were suspended in ultrapure MilliQ water (Type 1, Direct-Q 3UV, Merck-Millipore, Burlington, MA, USA) at pH 6 and in PBS (pH 7.3) in concentration 0.1 mg/mL. Absorption spectra were measured with a UV–visible spectrometer to confirm high absorption from UV up to 450 nm range. Structural phases of TiO₂ NP were confirmed by Raman spectra measurements with an α-SNOM Raman microspectrometer (Witec, Ulm, Germany), with excitation by an Ar ion laser with 488 nm wavelength.

For the rat red blood cell (RBC) samples, 3 mL of whole blood was withdrawn from the tails of Wistar rats and transferred into EDTA-covered tubes. The research methods were approved by the Animal Care and Use Committee of National Dong Hwa University (Approval ID100004). The rats were narcotized for the experiments. RBC samples were prepared and treated with NP as described previously [3,4]. In short, the RBC were separated from the fresh whole blood using centrifugation, repeatedly washed with standard phosphate buffer saline (pH 7.4), then RBC mass was diluted with PBS at the ratio of 5:1000 µL (RBC:PBS). The spectra of RBC in an oxygenated state were measured in ambient conditions and after nitrogen gas purging for deoxygenated RBC.

The TiO₂ powders were diluted in PBS at 33 mg/mL concentration. The obtained suspension was added to the RBC:PBS sample and coincubated for 1 h with the final concentration of TiO₂ ~20 µg/mL in the prepared sample. The prepared sample suspensions of RBC with TiO₂ powders and control RBC without TiO₂ were placed on a Si substrate for Raman measurements. The samples were at ambient atmospheric pressure, so RBC were oxygenated [3,4]. Raman spectra were measured and the selected Raman signal intensity mapping was performed with α-SNOM Raman microspectrometer (Witec, Ulm, Germany), with excitation by an Ar ion laser of 488 nm wavelength, objective ×50, power ~0.3 mW at the output, scanning time 150–220 s in ambient atmosphere conditions. Raman mapping of 15–20 RBCs from each sample were performed. Every mapping contained 24 × 24 spectra which were analyzed using software of Witec α-SNOM ScanCntr Spectroscopy Plus 1.34 (Witec, Ulm, Germany) and then with Origin 9.1 (OriginLab, Northampton, MA, USA).

3. Results

3.1. TiO₂ Particles Characterization

The size and ζ-potential of rutile TiO₂(r) and anatase TiO₂(a) phases NP suspended in ultrapure MilliQ water were measured. The ζ-potentials were found to be 15.5 ± 0.37 mV for anatase NP and 11.1 ± 0.36 mV for rutile. The size of TiO₂(r) at that condition showed a narrow distribution with a maximum near 154.4 ± 16.1 nm; TiO₂(a) revealed two fractions, one centered near size 136 nm and the other, characterizing the aggregates, with an average size of 448.1 ± 27.3 nm. When dissolved in PBS (pH 7.3), the measured values of ζ-potentials became negative and equal to −23.4 ± 2.0 and −25.2 ± 1.43 mV, respectively for the TiO₂(r) and anatase TiO₂(a). From the particle size distribution, the TiO₂ in PBS aggregated or agglomerated and formed a wide distribution with the hydrodynamic size up to one micron (data not shown); however, some crucially decreased fraction of small particles or small aggregates still exists. We deliberately did not remove the larger aggregates, as we wanted to examine the more practical application when RBC encounter all kinds of TiO₂ aggregates, especially in PBS. The particles’ agglomeration significantly increased in PBS in comparison with water, as was shown for TiO₂ NP in solutions with high ionic strength at pH 7.3 [25]. In such studies, the primary size of the TiO₂ NP, usually tens of nm, together with the crystallinity, played a role in the properties of agglomerates, and the properties inherent in nanoscale particles could also manifest at TiO₂ interaction with cells. The observed average hydrodynamic size of agglomerates reached 751.2 ± 48.8 nm (rutile) and 673 ± 29.0 nm (anatase). The agglomeration and changes in ζ-potential may have happened due to the influence of PBS components, as has been shown previously [26], where ions such as Cl⁻, Na⁺, and PO₄⁻³ from PBS can affect the particle surface properties and stabilization of TiO₂ NP suspension.

Typical Raman spectra of the studied TiO₂ powders are shown in Figure 1. Characteristic peaks were observed and agreed with literature values [27]. The spectra confirm the structure of two kinds of TiO₂ NP used in this work. These observed characteristic peaks were used as markers to localize TiO₂ in the Raman mappings of RBC at their interaction with TiO₂.

3.2. Raman Analysis of the RBC Oxygenation State

The red blood cells’ main component is Hb, the iron-containing oxygen transporting protein. Raman spectroscopy is used for studies of Hb and RBC; aside from its low-invasive nature, it allows for the simultaneous identification of oxygenated and deoxygenated states of Hb. Hemoglobin’s Raman spectrum is strong and complex but with fingerprints from heme (the active center of Hb) and protein units [21,22,24]. The spectra of oxygenated and deoxygenated RBC are shown in Figure 2. These spectra were measured from individual RBC with 488 nm wavelength laser excitation. Raman vibrations in different oxygenation states exhibited obvious changes in the oxidation state sensitive region (1350 to 1380 cm⁻¹, ν₄ band), spin state sensitive region (1500 to 1650 cm⁻¹, ν₁₉, ν₃₇, and ν₁₀ bands), and the C-H deformation region (1200 to 1250 cm⁻¹), followed by T ↔ R (tense–relaxed) transition in the Hb molecule in the oxygenation–deoxygenation process [22,28,29]. Based on the light absorption spectrum of oxyHb and deoxyHb, excitation can be selected close to the absorption bands to result in selective enhancement of resonance Raman peaks [30]. The Soret band maxima of oxy- and deoxy-Hb were 415 and 430 nm, respectively, and it has been shown that at 488 nm wavelength laser excitation, clear spectral changes in the 1350–1370 cm⁻¹ (ν₄ band) can be observed [28]. This band was assigned to the symmetrical pyrrole half-ring stretching vibration and the shift from 1376 cm⁻¹ to 1358 cm⁻¹ was demonstrated as a marker of deoxygenation at resonance Raman study of the oxygenation–deoxygenation process [30]; the peak’s intensity ratio was proportional to the portion of oxygenated Hb [4,30].

Typical spectra for oxyHb and deoxyHb are displayed in Figure 2, with the markers for the Raman mapping. The Raman mapping of RBC treated with TiO₂ was performed and compared with the control of the untreated RBC. Figure 3 shows the typical mapping of control RBC (Figure 3a) and RBC treated with TiO₂(r) (Figure 3b,c). Raman mapping illustrated the spatial distribution of Raman signal intensity along the sample; the distribution of oxyHb was mapped via Raman peak with a maximum in the 1365–1380 cm⁻¹ range (O in Figure 2) and distribution of deoxyHb via a peak in the 1350–1360 cm⁻¹ range (D in Figure 2). TiO₂ was visualized via mapping of Raman signal in the 600–650 cm^-1 range where A_1g band of Rutile (610 cm⁻¹) and E_g band of Anatase (639 cm^-1) were revealing, peaks shown in Figure 1. Note that the measurements were performed in ambient conditions; thus, the RBC should be mostly oxygenated. The spectra measured from the TiO₂-treated and untreated RBC allowed us to observe and discuss the effect of TiO₂ NP on oxygenated RBC, while mapping of the intensity of Raman signal in characteristic RBC (Hb) and TiO₂ ranges allowed, correspondingly, visualizing the relative distributions of oxygenated oxyHb (Figure 3-I), deoxygenated deoxyHb (Figure 3-II), and TiO₂ localization at interaction with the RBC (Figure 3b,c-III) in comparison with the control (Figure 3a-III). Note that TiO₂ is not fluorescent and cannot be easily observed with the methods of laser fluorescence scanning microscopy, while Raman mapping serves as an alternative means of TiO₂ visualization.

In the images of a control cell (Figure 3a) the part of oxygenated Hb was localized in the center of the RBC (Figure 3a-I), the relatively deoxygenated Hb was distributed on the peripheral part of RBC (Figure 3a-II). In general, a similar distribution was observed for mapping of RBCs treated with TiO₂. In Figure 3b,c, Raman mappings of an RBC treated with TiO₂(r) and TiO₂(a) are presented, and the maps of relatively oxygenated Hb (Figure 3b,c-I) and relatively deoxygenated Hb (Figure 3b,c-II) were added with a distribution of TiO₂ (Figure 3b,c-III). Well-detectable aggregates of TiO₂ localized on the outer surface of the RBC membrane were observed in both cases of using Anatase and Rutile particles. In spite of a strong Raman signal of TiO₂, we did not observe any NP which could be considered localized inside the cell.

Both Rutile and Anatase NP were tested before, and the toxicity of Rutile and Anatase could differ. In different studies, the results were controversial and depended significantly on the experimental conditions [31,32,33,34]. In order to discuss the possible effect of both kinds of TiO₂ NP on RBC, the spectra were analyzed focusing on the areas of maximum oxygenation (as shown in Figure 3-I) and maximum deoxygenation (as in Figure 3-II) in the Raman maps of each studied individual RBC. In Figure 4, the spectra of individual RBCs in a narrow range containing the band ν₄ are shown. A large number of individual cells was analyzed, the variability in Hb oxygenation state distribution along one cell and an average oxygenation degree for the different cells was observed. The spectra for each analyzed RBC were selected to demonstrate the highest (I) and lowest (II) oxygenation degree, which could be observed in the same individual cell (Figure 4a). The spectra in Figure 4a,b were measured from similar areas of two different cells and show the oxygenation degree of variability between cells. Figure 4(1) shows spectra for control untreated RBCs, while Figure 4(2,3) show spectra for RBCs treated with Rutile TiO₂(r) and TiO₂(a), respectively.

Figure 4 shows the data for cells indicating the largest difference found in the oxygenation degree inside one cell differs for untreated and TiO₂-treated RBC. The spectra shown for higher (I) and lower (II) oxygenation states of control RBC (Figure 4(1)) with the maximum at 1372 cm⁻¹ indicated mostly oxygenated Hb. Only a weakly expressed shoulder (arising from the peak at 1356 cm⁻¹) was observed in the spectra (II) both in Figure 4a(1),b(1), so the content of deoxygenated Hb can be considered very low. However, this low content of particularly deoxygenated Hb was still observable with Raman mapping in Figure 3a-II.

As the Raman peak intensity directly depends on the number of oscillators, the oxygenation degree SO₂ can be considered proportional to the ratio of intensities I₁₃₇₂ and I₁₃₅₆ [4,30], and can be expressed as SO₂ = A × I₁₃₇₂/(I₁₃₇₂ + I₁₃₅₆) + B, where I is the peak intensity at the corresponding wavenumber, and A and B are coefficients, which have to be determined via a calibration with independent oximetry measurements [28].

Figure 4a(1),b(1) show that the control RBC were predominantly oxygenated, as expected in ambient conditions. I₁₃₇₂/(I₁₃₇₂ + I₁₃₅₆) can be considered close to 1 in the RBC center (as in Figure 3a), and it decreased in the thin layer on the periphery in the cell (Figure 3b). Figure 4(2,3) show the spectra for RBC treated with different kinds of TiO₂. In contrast to Figure 4(1), in Figure 4a(2),a(3), oxyHb in the center of the cell, with a ratio I₁₃₇₂/(I₁₃₇₂+ I₁₃₅₆) close to 1, coexisted with significantly deoxygenated Hb on the cell periphery. An even lower oxygenation degree was estimated along the whole cells, presented by spectra in Figure 4b(2),b(3), where the central part of RBC contained only particularly oxygenated Hb, with both well-observable peaks at 1372 and 1356 cm⁻¹, while the periphery revealed mostly deoxyHb. Thus, the content of deoxyHb was observed to be higher in both TiO₂-treated cells than in the control. Spectra extended in the spin-sensitive range (Figure 5) also confirmed a decrease in oxyHb markers (ν₃₇ at 1588 cm⁻¹ and ν₁₀ at 1640 cm⁻¹) for TiO₂-treated RBC (Figure 5b,c) in comparison with untreated cells (Figure 5a). This means that TiO₂ adsorbed on the RBC membrane may affect Hb oxygenation degree in the air environment when RBC should be oxygenated. Analyzing the variability of TiO₂ effect on oxygenated RBC, no difference between the Rutile and Anatase particles was observed. Maxima for ν₄ in oxygenated and deoxygenated states were determined via the spectra deconvolution (Figure 6), and the value I₁₃₇₂/(I₁₃₇₂ + I₁₃₅₆) characterizing the variability of RBC oxygenation for untreated and treated RBC is presented in the inset of Figure 6.

4. Discussion

In previous studies, a shift of the ν₄ band in the process of Hb oxygenation was observed [30], and it was demonstrated that the ratio of intensities of the peaks at 1372 cm⁻¹ and 1356 cm⁻¹ could be used as a marker of the oxygenation state of Hb [10,21,30]. Additionally, in Raman mapping, we observed the inhomogeneous distribution of Hb with various oxygenation degree values along an individual RBC, and analyzed how the adsorption of TiO₂ on the RBC membrane affected this distribution. The predominant localization of oxygenated Hb is in the RBC center (Figure 3a-I), while a small fraction of partially deoxygenated Hb is distributed on the peripheral part of the control RBC (Figure 3a-II). It is consistent with the studies of deoxyHb binding on RBC membrane as a result of electrostatic interactions, covalent association with the membrane components by disulfide bonds, and adsorption to membrane lipids via hydrophobic interactions [35]. In particular, the reversible association of deoxyHb with the RBC membrane via interaction with band 3 protein provides an active role of the membrane in RBC gases’ transport function [35]. It is considered a “molecular switch” mechanism for regulating RBC biology (metabolic and transport processes, signal transduction, etc.) [36]. Similarly, RBCs treated with both kinds of TiO₂ demonstrated the peripheral localization of some portion of deoxyHb and the central localization of Hb with a higher oxygenation degree (Figure 3b, for TiO₂(r) and Figure 3c for TiO₂(a)). However, the oxygenation degree of the whole TiO₂-treated RBC was decreased as is shown in the inset of Figure 6.

As the plasma membrane of RBC plays a fundamental role in the oxygen (and other gases) transport by RBC and in maintaining RBC metabolism [37], the membrane structural changes, which can be created by adsorbed NP aggregates, can affect the RBC state [3,4,10,11,18]. Observed TiO₂ aggregates or agglomerates on the RBC membrane agreed, e.g., a previous study showed TiO₂ particles sticking to RBC membrane regardless of their surface charge [18]. It suggests the mostly nonelecrostatic nature of the NP–membrane adhesion interaction; however, it is not clear, and more detailed studies are needed to confirm. The membrane permeability for oxygen is determined by the membrane molecules’ mobility; thus, the interaction between the membrane and adhered NP aggregates or agglomerates can affect the conditions for the gas molecules’ transmembrane diffusion and for Hb adsorption on the membrane. Additionally, as was shown [10], the interaction with adsorbed NP may create an obstacle for the exchange of ions and polar molecules via protein channels and also due to production of reactive oxygen species [14] and lipid peroxidation. These reasons can lead to altering cell membrane properties, pH imbalances inside the cells, and finally can affect the oxygen binding affinity of Hb. As a result, similar deoxygenation under NP treatment has been observed [10] for RBC trapped with optical tweezers and treated with Ag or Au NP.

Additionally, using an excitation laser with 488 nm wavelength, we needed to consider the probability of laser effects directly on Hb [38], such as photodegradation of Hb (denaturation, hemichrome formation, and potential aggregation) in the process of Raman measurements. Some researchers suggest that the position of the ν₄ peak can indicate the iron ion transition to the ferric form [39], and this is considered as one of the markers of the Hb photodamage [40]. This effect is pronounced in the presence of both kinds of TiO₂ in comparison with control. Thus, this combined effect is facilitated by the oxidative stress due to photocatalytic and phototoxic properties of TiO₂ at the excitation close to the TiO₂ absorption range [41]. Most of the resulting Hb forms contained the ferric heme iron and could hardly be distinguished from oxyHb. However, comparison of the spectra in Figure 4 and Figure 5 shows that the position of the ν₄ peak at 1372 cm⁻¹ correlated with the presence of other peaks characterizing oxyHb (ν₃₇ 1588 cm⁻¹ and ν₁₀ 1640 cm⁻¹), while ν₄ observed at 1356 cm⁻¹ correlated with their decreasing. This decrease in intensity of ν₃₇ and ν₁₀ peaks allows us to conclude that particular deoxygenation of Hb takes place, rather than Hb photodegradation. We suggest TiO₂ affects Hb state through affecting the membrane properties and stimulats Hb deoxygenation.. We cannot exclude also the input of photodissociation of Hb, also resulting in the observation of higher content of deoxyHb. We suggest also that, together with influence on Hb through affecting the membrane properties and stimulating Hb deoxygenation membrane-associated TiO₂, aggregates/agglomerates can promote photo or photothermal dissociation of the oxyHb.

Use of TiO₂ NP for nanotherapy is currently widely discussed [13]. The limitations of their application as photosensitizers for photodynamic therapy are discussed in terms of possible side-effects, but they are considered nontoxic for their use in drug delivery or for ultrasound activation of ROS formation for anticancer treatment [42,43]. However, significant and various detrimental influences of TiO₂ NP on blood and blood components are demonstrated, and understanding these effects and studying these mechanisms, including the nonphotoactivated ones, is necessary.

In conclusion, the TiO₂ effect on RBC was observed, resulting in a decrease in the oxygenation degree of RBC. We suggest that TiO₂ affects Hb oxygenation state via altering RBC membrane properties. The disturbing treatment of the laser in the presence of TiO₂ is also probable, resulting in oxygen photodissociation of Hb. These effects should be considered in the development of the method of TiO₂ NP theranostic applications which imply NP interaction with blood.